16:00 〜 16:15
[MIS04-09] Assessment of possible landslide susceptibility under climate change : A case in Southwestern Taiwan
★Invited Papers
キーワード：climate change, landslide susceptibility, machine learning, Southwestern Taiwan
This study investigated the future spatiotemporal distribution of landslide susceptibility in the Nanzixian river watershed (NRW) in Southwestern Taiwan under four future climate change scenarios. This study explored the spatial and temporal distributions of landslide susceptibility in the NRW in Southwestern Taiwan under four future climate change scenarios. Ten relevant geomorphological factors were selected for constructing landslide susceptibility models by using the FR and LR methods, and the model with better performance was then selected for subsequent analysis. The landslide susceptibility model based on the FR method (AUC = 0.754, accuracy = 0.773) outperformed that based on the LR method (AUC = 0.732, accuracy = 0.759). Therefore, the landslide susceptibility model based on the FR method was used for subsequent analysis.
Past and future rainfall patterns in the NRW were compared in detail. The historical data used in this study were the daily rainfall data collected at two rainfall stations, namely the Jiaxian and Paiyun rainfall stations, from 2000 to 2023. The future rainfall data used in this study were downscaled daily rainfall data from 2024 to 2100 that were obtained from the TCCIP. No obvious difference in average annual rainfall was noted between the past and the future; however, significant differences were discovered in the maximum and minimum annual rainfall between the past and the future. Under the four considered climate scenarios, the future average annual rainfall in the NRW is expected to be 95.7%–107.6% of the past average annual rainfall in this watershed. Moreover, under the aforementioned scenarios, the future maximum and minimum annual rainfall values in the NRW are expected to be 97.9%–179.5% of the past maximum annual rainfall and 50.0%–105.0% of the past minimum annual rainfall in this watershed, respectively. The predicted cumulative rainfall during the rainy season (May to October) will exceed 95% of the annual rainfall in the future, which indicates that rainfall concentration will be excessive in the rainy season, leading to more floods and droughts in the NRW.
This study estimated past and future daily rainfall values with return periods of 5, 10, 25, 50, 100, and 200 years. Under the four climate change scenarios, no notable increase in daily rainfall is expected in the NRW. The ratio between the mid-future and past daily rainfall with return periods of 5, 10, 25, 50, 100, and 200 years is 1.30–1.51 under the SSP 3-7.0 scenario. In addition, the ratio between the far-future and past daily rainfall with return periods of 5, 10, 25, 50, 100, and 200 years is 1.08–1.40 under the four considered scenarios.
During Typhoon Morakot, the maximum daily rainfall in the NRW was 917.8 mm, which resulted in a major landslide disaster. The daily rainfall with a return period of 50–100 years will exceed 917.8 mm in the future under the four considered climate change scenarios. Thus, the intensity and frequency of extreme rainfall events in the NRW are expected to increase in the far future under climate change. This result indicates that major rainfall-induced landslide disasters will become more frequent in this watershed in the far future.
Areas with middle-high and high landslide susceptibility will be distributed in the upstream and midstream parts of the NRW. In the near future, the mean landslide susceptibility in the NRW is expected to exceed 6.9 (which was the mean landslide susceptibility in the NRW during Typhoon Morakot in 2009) with a return period of 200 years under the SSP 1-2.6 scenario and >100 years under the SSP 2-4.5, SSP 3-7.0, and SSP 5-8.5 scenarios.
Past and future rainfall patterns in the NRW were compared in detail. The historical data used in this study were the daily rainfall data collected at two rainfall stations, namely the Jiaxian and Paiyun rainfall stations, from 2000 to 2023. The future rainfall data used in this study were downscaled daily rainfall data from 2024 to 2100 that were obtained from the TCCIP. No obvious difference in average annual rainfall was noted between the past and the future; however, significant differences were discovered in the maximum and minimum annual rainfall between the past and the future. Under the four considered climate scenarios, the future average annual rainfall in the NRW is expected to be 95.7%–107.6% of the past average annual rainfall in this watershed. Moreover, under the aforementioned scenarios, the future maximum and minimum annual rainfall values in the NRW are expected to be 97.9%–179.5% of the past maximum annual rainfall and 50.0%–105.0% of the past minimum annual rainfall in this watershed, respectively. The predicted cumulative rainfall during the rainy season (May to October) will exceed 95% of the annual rainfall in the future, which indicates that rainfall concentration will be excessive in the rainy season, leading to more floods and droughts in the NRW.
This study estimated past and future daily rainfall values with return periods of 5, 10, 25, 50, 100, and 200 years. Under the four climate change scenarios, no notable increase in daily rainfall is expected in the NRW. The ratio between the mid-future and past daily rainfall with return periods of 5, 10, 25, 50, 100, and 200 years is 1.30–1.51 under the SSP 3-7.0 scenario. In addition, the ratio between the far-future and past daily rainfall with return periods of 5, 10, 25, 50, 100, and 200 years is 1.08–1.40 under the four considered scenarios.
During Typhoon Morakot, the maximum daily rainfall in the NRW was 917.8 mm, which resulted in a major landslide disaster. The daily rainfall with a return period of 50–100 years will exceed 917.8 mm in the future under the four considered climate change scenarios. Thus, the intensity and frequency of extreme rainfall events in the NRW are expected to increase in the far future under climate change. This result indicates that major rainfall-induced landslide disasters will become more frequent in this watershed in the far future.
Areas with middle-high and high landslide susceptibility will be distributed in the upstream and midstream parts of the NRW. In the near future, the mean landslide susceptibility in the NRW is expected to exceed 6.9 (which was the mean landslide susceptibility in the NRW during Typhoon Morakot in 2009) with a return period of 200 years under the SSP 1-2.6 scenario and >100 years under the SSP 2-4.5, SSP 3-7.0, and SSP 5-8.5 scenarios.